PDA

View Full Version : neutral density filters to remove moving objects



empkae
02-13-2010, 02:48 PM
I would like to understand how to use a very dark neutral density filter to remove moving objects. Example: A picture of a building where there are people milling about. If the filter is dark enough, and the people move enough, and the exposure is long enough, then most of the moving people will be reduced to phantom blurs, or gone entirely, leaving the building nice and sharp.

My problem is figuring out the math to get the right filter to do this.

And, is it possible to stack a couple of filters, one 1/3 as dark as needed above, and one 2/3 as dark as needed above, and get the same effect? Reason, to have some means to accommodate more levels of ambient light, from cloudy to bright sun. If this can be done, what are the "right" filters to use?

Mad Aussie
02-13-2010, 03:04 PM
You'll find the right ND filters to use will depend entirely on the scenes brightness and dynamic range. This would vary quite a bit I would think.
You can stack the ND filters but depending on your focal length you will have to be away of vignetting from the filters.
I've never heard of any particular formula/math to tell you what/how many ND filters to use though.

AntZ
02-13-2010, 05:59 PM
I'm guessing here, but I would think you work out what time you need to get the desired result to blur the moving objects, then what your desired DOF and therefore apperture needs to be. You can then adjust the filter darkness and ISO to get the desired result.

I have shot of my bother chasing a ball taken late one afternoon in which he is almost invisible. I had been playing with a CPL on to slow the water. It was less than 1sec exposure, but my brother was moving quite fast.

ArtTwisted
02-14-2010, 01:05 AM
to completely remove all people ? Im guessing go as dark as you can and see what happens, stack if you have too. Havent tried it though.

erwin.zeez
05-05-2010, 02:53 AM
to completely remove all people ? Im guessing go as dark as you can and see what happens, stack if you have too. Havent tried it though.

have no idea:sorry:

Greg_Nuspel
05-05-2010, 06:31 AM
Biggest problem with stacking is the introduction of more layers of glass/plastic in front of your lens. Each surface is a source of possible reflections, flare or in the case of low quality filters distortion.

Marko
05-05-2010, 08:47 AM
I've not done any tests on this at all but to my mind this has more to do with the ratio between shutter speed and the length of time any given person is standing in any one spot. Obviously if you increase the shutter speed through the use of an ND you have a better chance of people moving about to the point where they may blur or be gone....but how long will they stay, will they be wearing a bright white shirt, will they move front to back versus side to side...

But there is no easy math formula here. It all depends on the slowest moving person and you have zero control over that. You could have 1 person standing and eating a sandwich for 5 minutes....

you will likely have a much easier time with software.... I cannot come up with it now but by taking several shots of the same scene over time, there is software that can 'erase' the moving parts and get you your clear building. You could likely do this yourself by layering a bunch of shots over each other and erasing the people from 1 layer to the next.

ericmark
05-06-2010, 09:17 PM
I did one experiment years ago with 25ASA film with tripod on the rows in Chester with the cross as subject. Smallest aperture and two polarising filters turned to get lowest light I could. I think around 3 minutes but a long time ago. The result had just two people in picture in fact two police one male one female both A1 results so must have been standing very still.
And streets were full so yes it did work.
At that time there were no personal computers and software was not an option. However I have noted when combining images for panorama CS4 has managed not to put two arms in where people have moved between images so I would think it would be worth trying a stack process and see the results.